Insights
Insights
Frameworks, analysis, and perspectives on product analytics in the AI era.
Featured Framework
The Industrialization of Knowledge Work
A framework for understanding how AI changes organizations, not just productivity. Why the real advantage is building systems, not buying tools.
Read the full framework →Why NPS Is Overrated (And What to Measure Instead)
Net Promoter Score measures what people say, not what they do. Here are three metrics that actually predict retention and revenue, with the data to prove it.
Read more →ARPU Analysis: The Metric That Tells You If Your Pricing Is Broken
Average revenue per user is the most under-analyzed metric in SaaS. Here is how to calculate it, segment it, and use it to diagnose pricing problems before they compound.
Read more →Conversion Rate by Funnel Stage: Where Startups Actually Lose Users
Optimizing your overall conversion rate is a waste of time. Here is how to diagnose which funnel stage is broken, with benchmarks and a Python implementation.
Read more →Time to Value: The Onboarding Metric That Predicts Everything
The faster a user reaches their first value moment, the more likely they are to stay. Here is how to measure time to value, reduce it, and use it to predict retention.
Read more →DAU/MAU Ratio: What Your Stickiness Metric Actually Tells You (And What It Hides)
The DAU/MAU ratio is the simplest measure of product stickiness. It is also one of the most misinterpreted. Here is how to read it, benchmark it, and avoid the traps.
Read more →How to Measure Feature Adoption (And When to Kill a Feature)
Most teams ship features and never measure whether anyone uses them. Here is a framework for tracking adoption, identifying failures early, and making the kill decision with data.
Read more →How to Calculate Churn Rate for SaaS (And the 3 Ways Teams Get It Wrong)
Churn rate is simple to define and surprisingly easy to miscalculate. Here is the formula, three common mistakes, and a Python implementation that handles the edge cases.
Read more →Bank Statement Classification with AI: The Two-Layer Approach
Most firms still classify bank transactions by hand. A two-layer bookkeeping automation approach (keyword matching plus AI fallback) handles 99% automatically. Here is how it works and what to evaluate.
Read more →How to Calculate Customer Lifetime Value (And Why Most Startups Get It Wrong)
LTV is the most misused metric in startup analytics. Here is how to calculate it correctly, with the formula, common mistakes, and a Python snippet.
Read more →Experiment Velocity: The Metric That Separates Fast Companies from Slow Ones
The number of experiments your team runs per quarter predicts growth better than any single A/B test result. Here is how to measure and improve experiment velocity.
Read more →How to Choose a North Star Metric (A Framework That Actually Works)
Most teams pick the wrong North Star metric because they optimize for what is easy to measure, not what matters. Here is the framework I use with every client.
Read more →Retention Curve Analysis: The Metric That Decides Whether Your Product Survives
Retention curves tell you more about product-market fit than any other metric. Here is how to read them, analyze cohorts, and spot the signals that matter.
Read more →The Minimum Viable Experiment: A Framework for Early-Stage Rigor
Early-stage companies either skip experimentation entirely or replicate enterprise complexity. The middle ground requires only three components, and the most important one is the part teams skip.
Read more →North Star Metrics for AI Products: Why DAU Is the Wrong Choice
Traditional SaaS north star metrics reward login frequency. AI products need a metric that captures repeated value extraction, not repeated visits.
Read more →Will AI Replace Data Analysts? No. But It Will Expose Which Ones Were Misallocated.
AI eliminates the wrong kind of analyst work. What remains is the highest-value work most teams have been underinvesting in for years.
Read more →The Case for a Fractional Head of Data
Most growing companies need a data strategy, not another analyst. A fractional data leader builds the system in 90 days, then hands off execution.
Read more →Your A/B Test Didn't Fail. You Measured the Wrong Metric.
Most experiments fail not because the hypothesis was wrong, but because the team picked a metric that couldn't capture what they actually cared about.
Read more →The Decision Memo Template Your Data Team Needs (And Isn't Using)
The most effective analytical output is not a dashboard or a slide deck. It is a one-page decision memo with four sections.
Read more →Why Most AI Startups Get Their Activation Metric Wrong
The activation metric that worked for SaaS doesn't translate to AI products. Here's how to find the one that actually predicts retention.
Read more →Why Your Data Team's Best Analysis Never Drives Action
Every analytics team has a graveyard of Jupyter notebooks containing genuine insights nobody acted on. The gap between analysis and action is not technical. It is communicative.
Read more →Your AI Product's Retention Curve Is Lying to You
Blended retention curves hide the behavioral segments that matter most. Here's how to decompose them into actionable intelligence.
Read more →Why Dashboards Fail to Drive Decisions (And What to Build Instead)
A dashboard tells you what happened. A decision system tells you what to do about it. Most organizations are stuck on the wrong side of that line.
Read more →Data-Driven Is a Trap: Why the Most Popular Goal in Analytics Gets It Wrong
Data-driven sounds rigorous. It is also the wrong goal. The companies that make better decisions aim for something subtly but critically different.
Read more →Analytics Maturity Assessment: Every Company Overestimates. Here's Where You Actually Sit.
Most companies that call themselves data-driven sit at Level 2 of a five-level maturity framework. The gap to Level 4 is not technical. It is organizational.
Read more →